Hedge Fund Data

cake with hundreds and thousands and percentage sign candle

Constructing hedge fund indices is like baking a cake!

9 min read
Download Article

Seven (of the many) different processes and ingredients that can be used in constructing a hedge fund index. The choice between the processes and ingredients on offer can result in indices having wildly different results.

So what was the performance of the hedge fund industry in 2020?

  • A) 8.66%
  • B) 12.18%
  • C) 7.98%

Answer: All of the above AND more, it just depends on what methodology you use. The 2020 performance numbers above relate to asset weighted, equally weighted and median weighted returns respectively, calculated using Aurum’s Hedge Fund Data Engine1.

And there are more methodologies that could be introduced too. Compounding the differences that arise from the index methodology used, there are also huge variations in the constituents used by published hedge fund indices. For example, the constituents of the Aurum Hedge Fund Data Engine vary significantly from those of many well-established hedge fund database providers, that in turn differ from one another.

No wonder users of hedge fund industry indices scratch their heads at times when asking the question – ‘So what was the actual performance for X period?’ Understanding how an index is constructed is a critical factor in interpreting the data drawn from it.

It is for this reason that Aurum has always maintained its own database and indices, some of which have recently been made publically available.

Index methodologies

Source: Aurum Hedge Fund Data Engine. Industry Returns Jan 2017 – Jan 2021

Over the past year Aurum has been publishing data on hedge fund industry performance from its proprietary Hedge Fund Data Engine. The data published examines performance through a variety of dimensions like size, geography, strategy and liquidity and compares asset v equally weighted v median performance. Aurum has also published deep dives into specific hedge fund strategies and their sub-strategies.

The Aurum Hedge Fund Data Engine is, in a sense, just a continuation of the work that Aurum does as a long-standing, specialist hedge fund investor. After 26 years of investing in hedge funds, Aurum has developed and maintained a huge historical database on hedge fund performance. As far back as 1994, Aurum provided hedge fund database solutions to well-known industry data specialists.

The Hedge Fund Data Engine was the next step. We’ve developed a powerful analytical tool to examine performance and develop an array of hedge fund indices based on a wide breadth of dimensions (of which there are hundreds) for which each fund in the database has been tagged where possible. For instance a fund can be tagged by:

  • Strategy
  • Sub Strategy
  • Size
  • Liquidity Terms
  • Service Providers
  • Correlation to market indices
  • Management entity
  • Fee structure

to name but a few.

Part of the impetus for developing the Hedge Fund Data Engine was to address some of the problems Aurum has experienced when relying on externally available hedge fund indices for research. Aurum’s Hedge Fund Data Engine solves the following problems.

1. Industry indices don’t have data on many hard-closed and capacity constrained managers

Aurum has been an active allocator in the hedge fund space for over 26 years and understands that some managers don’t want their funds’ performance data to be freely available to any database subscriber.

Most database providers sell access to individual funds’ track records, but Aurum never discloses fund-level detail. Performance data in respect of a specific hedge fund is only ever conveyed in aggregated index performance numbers. For this reason, Aurum receives fund data that many database providers don’t.

Some indices only report ‘open’ hedge fund managers or investable assets. On the one hand, this can be because the database provider wants to report an “investable index”; on the other hand, it can be because sourcing data on closed funds isn’t possible for some index providers.

Aurum’s commingled fund of funds have approximately 50% of their AUM invested in closed hedge funds, and there is an increasing level of the industry ‘closed’ for new business. Aurum maintains ‘closed’ managers in its database where performance data is received.

Aurum is an allocator known to many of these closed managers. Hard-closed, capacity constrained, or more secretive managers may provide Aurum with data that they do not provide to ‘publically’ accessible database providers with whom they have no business reason to engage.  In that way, Aurum has an edge over many database providers who simply do not have access to the same information.

2. Industry indices can have incorrect strategy/sub-strategy classifications

Aurum’s analysts classify the funds within the database into relevant master and sub-strategies. Sometimes these differ from the underlying manager’s own classification.

The difference can arise for a number of reasons. For example, the individual(s) responsible for uploading fund data to a database may not fully understand the strategy of the fund in the context of the wider hedge fund industry; or, perhaps the manager may wish to represent themselves as running a particular strategy for marketing purposes.

Aurum will reclassify funds to the appropriate strategy, and equally may reclassify a fund again if, for example, the manager demonstrates style drift.

Many external database providers seek a competitive advantage based on the ‘size’ of their database. This could lead to funds that Aurum wouldn’t consider suitable for inclusion in ‘hedge fund’ datasets to be included. A significant proportion of some databases include long-only funds. Aurum strips out long-only funds – the database does include a bucket of long-biased managers, but this is distinct from long only managers that may represent their strategies as hedge funds.

3. Industry indices may lock down numbers, and so don’t always reflect a correct historic view of performance

Historic numbers in the Aurum hedge fund data engine are subject to change; performance is not “locked down”.

Whilst Aurum collects daily and weekly data on many hedge funds, the Hedge Fund Data Engine itself uses monthly data only. Depending on the manager, the timeliness of reporting can vary. Sometimes numbers are reported late (often several months after the NAV date, or even longer) and this can impact previously published Aurum monthly index numbers. Reported fund performance data are typically based on month end estimates, rather than ‘finals’. As such there can at times be reporting differences where a fund’s estimates have differed from finals. When final number are reported to Aurum’s database the finalised numbers are updated.

Historic performance numbers can also change as a result of Aurum reclassifying a hedge fund, as mentioned above, or because a new fund enters the database along with its previously unreported track record. Other databases may take different approaches, some may only include returns on a ‘going forward’ basis, as opposed to backfilling previously reported numbers. That said, we recognise issues with backfilling and the nature of survivorship bias potential here.

In some instances performance for a fund is updated for a particular month, but not AUM, and some funds only report AUMs on a quarterly basis. In such instances Aurum reserves the right to fill, or roll forward, AUMs as it considers appropriate. As and when updated AUMs are received these are reflected in the updated database.

Aurum looks to continually build and improve its database, and it is acknowledged that this can result in restating prior reported numbers. Other database providers often ‘lock down’ returns at month end. The advantage of this is that users are not subjected to revisions of data.

Aurum takes the approach that when the facts change, so does the data. Revision reflects a more up-to-date view of history as the evidence becomes available, rather than locking down at a convenient arbitrary point in time, i.e. 30 calendar day post month end.

4. Industry indices can be victims of errors reported by funds

Aurum will correct or suppress a fund’s reported numbers (AUM or performance) where necessary. Occasions where this is necessary include where an obvious ‘fat finger’ error has occurred. For example, if a manager has reported a $1bn AUM for many months, then suddenly reports a $10bn AUM one month, then back to a $1bn AUM the following month – in this situation Aurum will overwrite what it believes to be the incorrect data. Aurum may question fund track records directly with managers, or seek to cross-reference with other external databases. Other databases may not clean data for errors, placing reliance and onus on reporting managers to vet their own data.

5. Double counting AUM (e.g. reporting fund AUM in relation to multiple share classes)

Occasionally onshore and offshore funds, or different share classes are reported all listing the total AUM of the master fund, which duplicates AUM. In these cases Aurum uses the aggregated AUM, without double counting performance (which would impact equally weighted indices) and applies the performance of either: the lead series, the largest share class or the class with the longest track record. Other databases may include multiple share classes of the same fund whilst applying the same master fund AUM, therefore often double/treble (or worse) counting of AUMs. This can distort both asset weighted and equally weighted performance numbers and strategy and industry AUMs.

6. Many industry providers only publish one, or at most, two index weightings.

Aurum maintains three main indices types (Asset Weighted, Equally Weighted and Median Weighted) along with various other percentile/decile related indices.

Median (or percentile) weighted

The performance of the 50th percentile fund is taken as the representative performance for the index. All performance values from funds reporting above or below the 50th percentile are ignored.

Equal weighted (no Windsorization) v asset weighted

All reported funds have an equal weighting in respect to sizing in the index, regardless of AUM. As such a recently launched $10m fund has the same impact on the performance of the index as the world’s largest hedge fund.

Some industry observers prefer looking at asset weighted returns given they better reflect how the wider investment community has performed on actual deployed assets. That said there are those who would argue that asset weighting over-represents those managers who have successfully marketed their products or taken in too much money, as opposed to those more considered managers who deliberately restricted inflows of capital.

Aurum respects both arguments and considers all three major indices on their own merits and importantly looks to understand the difference between them and what it says about the underlying constituents. Asset weighted indices can be severely impacted by single large funds, i.e. Bridgewater, whereas under equally weighted models small funds that largely aren’t invested in by external capital gain equal prominence to funds where real, significant external capital has been deployed and verified.

It should be noted that Aurum’s hedge fund data engine does not engage in Windsorization of its equally weighted indices as standard (although we have the capability to). Windsorization is a systematic process where the extreme tails are cut out of the performance calculation, for instance at the 5th and 95th percentile. This process can reduce the impact of outliers on an index, especially within equally weighted indices. Whilst Aurum does not tend to Windsorize, it does apply the right to individually suppress or alter a fund’s reported number where it believes that an error has occurred in a manager’s reporting. In this manner Aurum can remove ‘extremes’ where it believes these to be error related. In 2020 Aurum’s equally weighted indices were skewed in part by some crypto currency funds which had a stellar year given the performance in the underlying assets.

7. Many industry index providers don’t provide net flow and P&L reporting

Whilst the Aurum Hedge Fund Data Engine uses reported AUM and performance numbers, assumptions are made with regard to the dollar P&L and net flow movements calculated. Dollar P&L is simply calculated based on a fund’s starting AUM multiplied by the monthly net performance return. Net flow data is then calculated by acknowledging the difference from the subsequent AUM and the prior months AUM with reference to the Dollar P&L. For instance a fund reports a month 1 AUM of $100m and a month 2 AUM of $120m. If reported performance for month 1 was 10%, when Aurum calculates Dollar P&L as $10m and therefore can assume that the net difference relates to net flows into the fund at the end of month 1, i.e. +$10m.

When designing the  hedge fund data engine, Aurum has worked to redress some of the issues that the Aurum due diligence teams have experienced using external hedge fund indices. 26 years+ experience allocating to hedge funds, supported by Aurum’s proprietary technology systems have facilitated the collection of a wealth of data on the hedge fund industry. Aurum is in a fantastic position as a long-standing allocator to hedge funds to not only access and collate data about the hedge fund industry, but also to understand nuances and weaknesses in the data.

This wealth of data creates insight into hedge fund performance that supports Aurum’s research and helps investors to develop a broader and deeper knowledge of the hedge fund industry.

You may also like

Macro hedge fund primer: uncovering the unconstrained


Global macro   |   Fixed income relative value   |   Macro emerging markets   |   CommoditiesIn summary Macro funds typically take positions…

Monthly hedge fund industry performance review – October 2023


Hedge fund performance was moderately negative in October; the average asset weighted hedge fund net return across all strategies was -0.34%. Most hedge…

Aurum’s quarterly review – Q3 2023


Performance for Aurum’s commingled fund of hedge funds $US classes ranged from +1.7% to +2.4% in the third quarter of 2023, against a backdrop of negative…

Hedge fund industry performance deep dive – Q3 2023


In summary… Five-year CAR for hedge funds was at 4.9% at the end of Q3, above bonds at -1.7% and just above equities at 4.1%. Global equities*** and…

Monthly hedge fund performance review – September 2023


Hedge fund industry performance was moderately positive in September; the average hedge fund return across all strategies was 0.08% (asset weighted net…

Monthly hedge fund performance review – August 2023


Hedge fund performance was moderately positive in August; the average hedge fund net return across all strategies was 0.16%. Most hedge fund strategy groups…

Monthly hedge fund performance review – July 2023


Hedge fund performance was generally positive in July; the average hedge fund net return across all strategies was 1.05%. All hedge fund strategy groups…

Hedge fund industry performance deep dive – H1 2023


In summary… The hedge fund industry* was up 3.4% in H1 23 with performance being heavily weighted to the start of Q1 and the end of Q2. The best performing…

Aurum’s quarterly review – Q2 2023


Performance for Aurum’s commingled fund of hedge funds $US classes ranged from -0.9% to +0.2% in the second quarter of 2023. The outlook at the beginning…

Monthly hedge fund performance review – June 2023


Hedge fund performance was generally positive in June; the average hedge fund net return across all strategies was 1.40%. Equity markets rallied in June,…

Quant hedge fund primer: demystifying quantitative strategies


QEMN   |   Statistical arbitrage   |   Managed futures/CTAs   |   Quant macro/GAA   |   Alternative risk premiaIn summary Quantitative…

Monthly hedge fund performance review – May 2023


Hedge fund performance was flat in May; the average hedge fund net return across all strategies was 0.02%. Market volatility was notably higher than observed…